DPDK  16.04.0
rte_ip_frag.h
Go to the documentation of this file.
1 /*-
2  * BSD LICENSE
3  *
4  * Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  *
11  * * Redistributions of source code must retain the above copyright
12  * notice, this list of conditions and the following disclaimer.
13  * * Redistributions in binary form must reproduce the above copyright
14  * notice, this list of conditions and the following disclaimer in
15  * the documentation and/or other materials provided with the
16  * distribution.
17  * * Neither the name of Intel Corporation nor the names of its
18  * contributors may be used to endorse or promote products derived
19  * from this software without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22  * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
24  * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
25  * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
26  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
27  * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
28  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
29  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
30  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
31  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
32  */
33 
34 #ifndef _RTE_IP_FRAG_H_
35 #define _RTE_IP_FRAG_H_
36 
44 #ifdef __cplusplus
45 extern "C" {
46 #endif
47 
48 #include <stdint.h>
49 #include <stdio.h>
50 
51 #include <rte_malloc.h>
52 #include <rte_memory.h>
53 #include <rte_ip.h>
54 #include <rte_byteorder.h>
55 
56 struct rte_mbuf;
57 
58 enum {
62  IP_MAX_FRAG_NUM = RTE_LIBRTE_IP_FRAG_MAX_FRAG,
64 };
65 
67 struct ip_frag {
68  uint16_t ofs;
69  uint16_t len;
70  struct rte_mbuf *mb;
71 };
72 
74 struct ip_frag_key {
75  uint64_t src_dst[4];
76  uint32_t id;
77  uint32_t key_len;
78 };
79 
80 /*
81  * @internal Fragmented packet to reassemble.
82  * First two entries in the frags[] array are for the last and first fragments.
83  */
84 struct ip_frag_pkt {
85  TAILQ_ENTRY(ip_frag_pkt) lru;
86  struct ip_frag_key key;
87  uint64_t start;
88  uint32_t total_size;
89  uint32_t frag_size;
90  uint32_t last_idx;
91  struct ip_frag frags[IP_MAX_FRAG_NUM];
93 
94 #define IP_FRAG_DEATH_ROW_LEN 32
97 struct rte_ip_frag_death_row {
98  uint32_t cnt;
99  struct rte_mbuf *row[IP_FRAG_DEATH_ROW_LEN * (IP_MAX_FRAG_NUM + 1)];
101 };
102 
103 TAILQ_HEAD(ip_pkt_list, ip_frag_pkt);
107  uint64_t find_num;
108  uint64_t add_num;
109  uint64_t del_num;
110  uint64_t reuse_num;
111  uint64_t fail_total;
112  uint64_t fail_nospace;
114 
117  uint64_t max_cycles;
118  uint32_t entry_mask;
119  uint32_t max_entries;
120  uint32_t use_entries;
121  uint32_t bucket_entries;
122  uint32_t nb_entries;
123  uint32_t nb_buckets;
124  struct ip_frag_pkt *last;
125  struct ip_pkt_list lru;
127  struct ip_frag_pkt pkt[0];
128 };
129 
131 #define RTE_IPV6_EHDR_MF_SHIFT 0
132 #define RTE_IPV6_EHDR_MF_MASK 1
133 #define RTE_IPV6_EHDR_FO_SHIFT 3
134 #define RTE_IPV6_EHDR_FO_MASK (~((1 << RTE_IPV6_EHDR_FO_SHIFT) - 1))
135 
136 #define RTE_IPV6_FRAG_USED_MASK \
137  (RTE_IPV6_EHDR_MF_MASK | RTE_IPV6_EHDR_FO_MASK)
138 
139 #define RTE_IPV6_GET_MF(x) ((x) & RTE_IPV6_EHDR_MF_MASK)
140 #define RTE_IPV6_GET_FO(x) ((x) >> RTE_IPV6_EHDR_FO_SHIFT)
141 
142 #define RTE_IPV6_SET_FRAG_DATA(fo, mf) \
143  (((fo) & RTE_IPV6_EHDR_FO_MASK) | ((mf) & RTE_IPV6_EHDR_MF_MASK))
144 
145 struct ipv6_extension_fragment {
146  uint8_t next_header;
147  uint8_t reserved;
148  uint16_t frag_data;
149  uint32_t id;
150 } __attribute__((__packed__));
151 
152 
153 
154 /*
155  * Create a new IP fragmentation table.
156  *
157  * @param bucket_num
158  * Number of buckets in the hash table.
159  * @param bucket_entries
160  * Number of entries per bucket (e.g. hash associativity).
161  * Should be power of two.
162  * @param max_entries
163  * Maximum number of entries that could be stored in the table.
164  * The value should be less or equal then bucket_num * bucket_entries.
165  * @param max_cycles
166  * Maximum TTL in cycles for each fragmented packet.
167  * @param socket_id
168  * The *socket_id* argument is the socket identifier in the case of
169  * NUMA. The value can be *SOCKET_ID_ANY* if there is no NUMA constraints.
170  * @return
171  * The pointer to the new allocated fragmentation table, on success. NULL on error.
172  */
173 struct rte_ip_frag_tbl * rte_ip_frag_table_create(uint32_t bucket_num,
174  uint32_t bucket_entries, uint32_t max_entries,
175  uint64_t max_cycles, int socket_id);
176 
177 /*
178  * Free allocated IP fragmentation table.
179  *
180  * @param btl
181  * Fragmentation table to free.
182  */
183 static inline void
184 rte_ip_frag_table_destroy( struct rte_ip_frag_tbl *tbl)
185 {
186  rte_free(tbl);
187 }
188 
210 int32_t
211 rte_ipv6_fragment_packet(struct rte_mbuf *pkt_in,
212  struct rte_mbuf **pkts_out,
213  uint16_t nb_pkts_out,
214  uint16_t mtu_size,
215  struct rte_mempool *pool_direct,
216  struct rte_mempool *pool_indirect);
217 
218 /*
219  * This function implements reassembly of fragmented IPv6 packets.
220  * Incoming mbuf should have its l2_len/l3_len fields setup correctly.
221  *
222  * @param tbl
223  * Table where to lookup/add the fragmented packet.
224  * @param dr
225  * Death row to free buffers to
226  * @param mb
227  * Incoming mbuf with IPv6 fragment.
228  * @param tms
229  * Fragment arrival timestamp.
230  * @param ip_hdr
231  * Pointer to the IPv6 header.
232  * @param frag_hdr
233  * Pointer to the IPv6 fragment extension header.
234  * @return
235  * Pointer to mbuf for reassembled packet, or NULL if:
236  * - an error occured.
237  * - not all fragments of the packet are collected yet.
238  */
239 struct rte_mbuf *rte_ipv6_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
240  struct rte_ip_frag_death_row *dr,
241  struct rte_mbuf *mb, uint64_t tms, struct ipv6_hdr *ip_hdr,
242  struct ipv6_extension_fragment *frag_hdr);
243 
244 /*
245  * Return a pointer to the packet's fragment header, if found.
246  * It only looks at the extension header that's right after the fixed IPv6
247  * header, and doesn't follow the whole chain of extension headers.
248  *
249  * @param hdr
250  * Pointer to the IPv6 header.
251  * @return
252  * Pointer to the IPv6 fragment extension header, or NULL if it's not
253  * present.
254  */
255 static inline struct ipv6_extension_fragment *
256 rte_ipv6_frag_get_ipv6_fragment_header(struct ipv6_hdr *hdr)
257 {
258  if (hdr->proto == IPPROTO_FRAGMENT) {
259  return (struct ipv6_extension_fragment *) ++hdr;
260  }
261  else
262  return NULL;
263 }
264 
288 int32_t rte_ipv4_fragment_packet(struct rte_mbuf *pkt_in,
289  struct rte_mbuf **pkts_out,
290  uint16_t nb_pkts_out, uint16_t mtu_size,
291  struct rte_mempool *pool_direct,
292  struct rte_mempool *pool_indirect);
293 
294 /*
295  * This function implements reassembly of fragmented IPv4 packets.
296  * Incoming mbufs should have its l2_len/l3_len fields setup correclty.
297  *
298  * @param tbl
299  * Table where to lookup/add the fragmented packet.
300  * @param dr
301  * Death row to free buffers to
302  * @param mb
303  * Incoming mbuf with IPv4 fragment.
304  * @param tms
305  * Fragment arrival timestamp.
306  * @param ip_hdr
307  * Pointer to the IPV4 header inside the fragment.
308  * @return
309  * Pointer to mbuf for reassebled packet, or NULL if:
310  * - an error occured.
311  * - not all fragments of the packet are collected yet.
312  */
313 struct rte_mbuf * rte_ipv4_frag_reassemble_packet(struct rte_ip_frag_tbl *tbl,
314  struct rte_ip_frag_death_row *dr,
315  struct rte_mbuf *mb, uint64_t tms, struct ipv4_hdr *ip_hdr);
316 
317 /*
318  * Check if the IPv4 packet is fragmented
319  *
320  * @param hdr
321  * IPv4 header of the packet
322  * @return
323  * 1 if fragmented, 0 if not fragmented
324  */
325 static inline int
326 rte_ipv4_frag_pkt_is_fragmented(const struct ipv4_hdr * hdr) {
327  uint16_t flag_offset, ip_flag, ip_ofs;
328 
329  flag_offset = rte_be_to_cpu_16(hdr->fragment_offset);
330  ip_ofs = (uint16_t)(flag_offset & IPV4_HDR_OFFSET_MASK);
331  ip_flag = (uint16_t)(flag_offset & IPV4_HDR_MF_FLAG);
332 
333  return ip_flag != 0 || ip_ofs != 0;
334 }
335 
336 /*
337  * Free mbufs on a given death row.
338  *
339  * @param dr
340  * Death row to free mbufs in.
341  * @param prefetch
342  * How many buffers to prefetch before freeing.
343  */
344 void rte_ip_frag_free_death_row(struct rte_ip_frag_death_row *dr,
345  uint32_t prefetch);
346 
347 
348 /*
349  * Dump fragmentation table statistics to file.
350  *
351  * @param f
352  * File to dump statistics to
353  * @param tbl
354  * Fragmentation table to dump statistics from
355  */
356 void
357 rte_ip_frag_table_statistics_dump(FILE * f, const struct rte_ip_frag_tbl *tbl);
358 
359 #ifdef __cplusplus
360 }
361 #endif
362 
363 #endif /* _RTE_IP_FRAG_H_ */